On Lying and Being Lied To: A Linguistic Analysis of Deception in Computer-Mediated Communication

This study investigated changes in both the liar's and the conversational partner's linguistic style across truthful and deceptive dyadic communication in a synchronous text-based setting. An analysis of 242 transcripts revealed that liars produced more words, more sense-based words (e.g., seeing, touching), and used fewer self-oriented but more other-oriented pronouns when lying than when telling the truth. In addition, motivated liars avoided causal terms when lying, whereas unmotivated liars tended to increase their use of negations. Conversational partners also changed their behavior during deceptive conversations, despite being blind to the deception manipulation. Partners asked more questions with shorter sentences when they were being deceived, and matched the liar's linguistic style along several dimensions. The linguistic patterns in both the liar and the partner's language use were not related to deception detection, suggesting that partners were unable to use this linguistic information to improve their deception detection accuracy.